Releases: koaning/embetter
Releases · koaning/embetter
0.6.4
We now support faster inference on M1 macs natively out of the box!
0.6.2
Add direct support for matrouska models and make sentence-transformers a core dependency.
0.6.1
Fixed a keras import bug.
0.6.0
This version introduces breaking changes by splitting the finetuning submodule into tuners
and learners
. These are now the available components.
# Finetuning components
from embetter.finetune import FeedForwardTuner, ContrastiveTuner, ContrastiveLearner, SbertLearner
The docs also got a nice update as a result. We hope folks enjoy and enjoy experimenting!
0.5.3
Add support for lite text embedding models.
0.5.2
Added support for a difference model.
0.5.1
New documentation pages and support for Gensim/KerasNLP.
0.5.0
Introduced a breaking change: external embedding providers will now require an environment key. Feels safer to force users to use that.
0.4.1
Added a whole bunch of utilities #71.
0.4.0
Add support for caching the embeddings.